Lecture 7: Incremental (Small-Signal) Analysis

What this lecture covers

The previous lecture ended with a working but badly distorted music-over-light-beam circuit, because the LED-like device used had an exponential, nonlinear relationship between current and voltage. This lecture answers how to get a linear response out of that same nonlinear device without changing it: bias the device with a large DC offset so it operates at a fixed point, then superimpose a small time-varying signal on top of that offset.

By the end, you understand the small-signal (incremental) method both intuitively, a small enough piece of any curve looks like a straight line, and mathematically, via a Taylor series expansion that shows the incremental output is proportional to the incremental input when the input excursion is small. This method, and the small-signal resistance it produces, becomes a recurring tool for analyzing nonlinear devices like diodes and transistors later in the course.

Key ideas

• Local linearity: any smooth nonlinear curve looks approximately like a straight line if you zoom in closely enough around a single point.
• DC offset (bias point): choosing an operating point VD (and corresponding ID) around which the device will operate, shifting the signal into a region of the curve of interest.
• Small-signal superposition: adding a small time-varying signal vd on top of the DC offset VD, so the total instantaneous voltage is VD + vd, with the total current similarly ID + id.
• "Boost and shrink": the two-part trick of biasing the signal up (DC offset) and scaling it down (small amplitude) so the excursions stay within the locally linear region of the curve.
• Taylor series justification: expanding ID = F(VD) around the DC bias point and dropping second-order and higher terms shows that the incremental output id is approximately dF/dVD (evaluated at the bias point) times the incremental input vd, a linear relationship.
• Small-signal resistance: for the lecture's exponential device (ID = A * e^(B*VD)), the derivative works out to id = ID * B * vd, which has the same form as Ohm's law, id = vd / Rd, where Rd = 1 / (ID * B).
• Small-signal equivalent circuit: once a nonlinear device is biased and its small-signal resistance found, it can be replaced by that resistor for the purpose of analyzing the small-signal (incremental) response, turning a nonlinear problem back into a linear one.

Walkthrough

Review: where the playground stands (2:03)

The lecture recaps the overall structure covered so far: the lumped circuit playground, its linear subset (where superposition and Thevenin apply), the nonlinear region, and the digital region (which is nonlinear overall but linear for a fixed set of switch inputs). It reintroduces the light-emitting exponential device from last time, replaying the distorted music demo as a reminder of the problem to be solved.

Class discussion and Zen (14:25)

Several student suggestions are considered and set aside, including adding a compensating element or digitizing the signal first, as overkill. The lecture leads students toward the answer through an analogy to Zen focus: zooming in on a small enough region of the device's nonlinear curve, that region looks approximately like a straight line, which is the key insight behind the solution.

The "boost and shrink" trick, demonstrated (17:27)

The lecture explains the fix intuitively: add a DC offset to bias the input signal into a chosen operating region, and shrink the amplitude of the superimposed signal so its excursions stay confined to a small, locally linear stretch of the curve. A live demonstration shows the distorted signal becoming progressively cleaner as the input is first shrunk and then boosted with a DC offset, and the same improvement is heard directly when music is played through the biased circuit.

Formalizing the small-signal method (25:38)

The method is named small-signal analysis, incremental analysis, or the small-signal discipline, and stated in three steps: operate at a DC bias point, superimpose a small signal on top of it, and observe that the resulting small-signal response is approximately linear. Notation is introduced: total signal equals DC offset plus small signal (for example, ID total = capital VD offset plus lowercase vd incremental), sometimes written with a delta symbol to emphasize the incremental change.

Mathematical derivation via Taylor series (32:58)

Starting from ID = F(VD), the lecture substitutes VD = (bias point) + delta VD and expands F in a Taylor series around the bias point. Dropping second-order and higher terms, valid because delta VD is small by design, leaves a linear relationship: the incremental output delta ID equals the derivative of F at the bias point times delta VD. A graphical interpretation shows this as approximating a point on the curve using the tangent line's slope at the bias point.

Applying the result to the exponential device and defining small-signal resistance (44:34)

Plugging the exponential device's relationship ID = A * e^(B*VD) into the Taylor result gives id = ID * B * vd, matching the form of Ohm's law. This defines a small-signal resistance Rd = 1 / (ID * B), which depends on the chosen bias point. The lecture closes by showing that, for small-signal purposes, the nonlinear device can be replaced by this resistor in a small-signal equivalent circuit, turning nonlinear analysis back into familiar linear circuit analysis, a technique the course will revisit in more detail in later lectures on diodes and transistors.

Before you watch

• Watch Lecture 6 first; this lecture directly continues its nonlinear analysis and the distorted music-over-light-beam demonstration.
• Familiarity with Taylor series expansion from calculus is needed to follow the mathematical derivation of the small-signal approximation.

Check your understanding

1. Why does biasing a nonlinear device with a DC offset and adding only a small signal on top produce an approximately linear response?
2. In the Taylor series derivation, which terms are dropped, and why is dropping them justified when delta VD is small?
3. How is the small-signal resistance Rd derived for the exponential device, and why does it depend on the chosen bias point ID?
4. What does it mean to build a "small-signal equivalent circuit," and why is this useful for analyzing circuits containing nonlinear devices?